A title review material auditing method, device, terminal and storage medium

Abstract

The application provides a title review material auditing method, device, terminal and storage medium, and belongs to the technical field of auditing and authentication. The method comprises the following steps: receiving a plurality of review related information and corresponding title review materials; generating a first prompt word according to all the review related information; preprocessing all the title review materials to obtain a plurality of pictures, inputting all the pictures and the first prompt word into a multimodal large model to obtain a first judgment conclusion; extracting text from all the pictures, inputting the text and the first prompt word into a semantic large model to obtain a second judgment conclusion; and determining a final conclusion about the relevance of the review related information and the title review materials based on the first judgment conclusion and the second judgment conclusion and outputting the final conclusion. The application can effectively improve the auditing efficiency by using two models to judge the relevance of the review related information and the title review materials respectively and determining the relevance of the review related information and the title review materials based on the output results of the two models.

Description

Technical Field

[0001] The present invention relates to the field of audit and certification technology, and in particular to a method, device, terminal and storage medium for auditing professional title evaluation materials. Background Art

[0002] In existing technology, applicants for professional title evaluations are required to select a series of supporting materials, including papers, patents, projects, awards, and so on, on the application interface and upload the corresponding documents. However, there is currently no effective way to ensure that the information entered by applicants on the form is completely consistent with the uploaded attachments, and manual verification is still required. This manual verification method is inefficient.

[0003] Therefore, the existing technology has defects and needs to be improved and developed. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method, device, terminal and storage medium for reviewing professional title evaluation materials in response to the above-mentioned defects of the prior art, aiming to solve the problem of low review efficiency of manual verification in the prior art.

[0005] The technical solutions adopted by the present invention to solve the technical problems are as follows:

[0006] In a first aspect, an embodiment of the present invention provides a method for reviewing professional title evaluation materials, the method comprising:

[0007] Receive several evaluation-related information input by the user in the professional title evaluation system interface, as well as the corresponding professional title evaluation materials uploaded by the user for each evaluation-related information;

[0008] Generate a first prompt word based on all the review-related information;

[0009] Preprocessing all of the professional title review materials to obtain a plurality of images, inputting all of the images and the first prompt word into a multimodal large model, and processing the multimodal large model to output a first judgment conclusion on the relevance of the review-related information and the professional title review materials;

[0010] After extracting text from all the images, inputting the text and the first prompt word into a semantic big model, and processing the semantic big model to output a second judgment conclusion on the relevance of the review-related information and the professional title review materials;

[0011] Based on the first judgment conclusion and the second judgment conclusion, a final conclusion on the relevance of the review-related information and the professional title review materials is determined and output.

[0012] In one embodiment, each review-related information includes the type of professional title review materials and the corresponding inspection items; generating the first prompt word based on all the review-related information includes:

[0013] Pre-set the corresponding inspection items for each type of professional title review materials, and configure the corresponding prompt word template for each inspection item;

[0014] Determine all corresponding inspection items according to the type of the professional title evaluation materials input by the user;

[0015] Obtain the prompt word template corresponding to each inspection item, and fill the relevant inspection item content entered by the user into the corresponding prompt word template;

[0016] When all the inspection items are filled in the corresponding prompt word template, a first prompt word is generated based on the filled prompt word template.

[0017] In one embodiment, all the professional title evaluation materials are pre-processed to obtain a number of images, including:

[0018] Determine the format of all the professional title evaluation materials and distinguish between image format files and non-image format files;

[0019] For professional title evaluation materials that are not in image format, convert them into images;

[0020] For professional title evaluation materials in image format, determine whether their resolution reaches the preset resolution;

[0021] If the preset resolution is not reached, a re-upload instruction including a description of the resolution requirement is generated, and a new image submitted by the user in response to the re-upload instruction is received, and the determination of whether the resolution of the new image reaches the preset resolution is continued until an image that reaches the preset resolution is obtained;

[0022] The pictures that have reached the preset resolution and the pictures that have undergone format conversion are combined to form the plurality of pictures.

[0023] In one embodiment, for professional title evaluation materials that are not in image format, converting them into images includes:

[0024] Determining whether the number of pages of each non-image format professional title evaluation material exceeds a preset number of pages;

[0025] If the number of pages of the professional title evaluation materials in non-picture format does not exceed the preset number of pages, the format of the materials is converted to obtain the corresponding pictures;

[0026] If the number of pages of the professional title evaluation materials in non-picture format exceeds the preset number of pages, they will be cut according to the preset number of pages, and the cut professional title evaluation materials in non-picture format will be format-converted to obtain corresponding pictures.

[0027] In one embodiment, based on the first judgment conclusion and the second judgment conclusion, determining and outputting a final conclusion on the relevance of the review-related information and the professional title review materials includes:

[0028] Determining whether the first judgment conclusion and the second judgment conclusion are consistent;

[0029] If the two are consistent, the first judgment conclusion or the second judgment conclusion is taken as the final conclusion and output;

[0030] If the two are inconsistent, a second prompt word is constructed, the picture and the second prompt word are input into the multimodal large model, and the final conclusion is output after processing.

[0031] In one embodiment, the multimodal big model outputs a first reasoning process while outputting a first judgment conclusion. The first reasoning process is the process of the multimodal big model reasoning the relevance between the review-related information and the professional title review materials. The semantic big model outputs a second reasoning process while outputting a second judgment conclusion. The second reasoning process is the process of the semantic big model reasoning the relevance between the review-related information and the professional title review materials.

[0032] In one embodiment, the step of constructing the second prompt word includes:

[0033] Obtaining a preset second prompt word template, where the second prompt word template is used for the multimodal large model to perform a two-choice judgment;

[0034] The first prompt word, the first judgment conclusion, the first reasoning process, the second judgment conclusion and the second reasoning process are filled into the second prompt word template to obtain a second prompt word.

[0035] In a second aspect, an embodiment of the present invention further provides a device for reviewing professional title evaluation materials, comprising:

[0036] The data receiving module is used to receive several review-related information input by the user in the professional title review system interface, as well as the corresponding professional title review materials uploaded by the user for each review-related information;

[0037] A prompt word generating module, configured to generate a first prompt word based on all the review-related information;

[0038] A first judgment module is configured to pre-process all of the professional title evaluation materials to obtain a plurality of images, input all of the images and the first prompt word into a multimodal large model, and output a first judgment conclusion on the relevance of the evaluation-related information and the professional title evaluation materials after processing by the multimodal large model;

[0039] A second judgment module is configured to extract text from all the images, input the text and the first prompt word into a semantic large model, process the semantic large model, and output a second judgment conclusion on the relevance of the review-related information and the professional title review materials;

[0040] The final conclusion generation module is used to determine and output the final conclusion on the relevance of the review-related information and the professional title review materials based on the first judgment conclusion and the second judgment conclusion.

[0041] In the third aspect, an embodiment of the present invention also provides a terminal, which includes: a memory, a processor, and a professional title evaluation material review program stored on the memory and runnable on the processor, and when the professional title evaluation material review program is executed by the processor, the steps of the professional title evaluation material review method as described above are implemented.

[0042] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores a professional title evaluation materials review program, and the professional title evaluation materials review program can be executed to implement the steps of the professional title evaluation materials review method as described above.

[0043] The beneficial effects of the present invention are as follows: the present invention receives a number of review-related information and corresponding professional title review materials; generates a first prompt word based on all of the review-related information; pre-processes all of the professional title review materials to obtain a number of images, inputs all of the images and the first prompt word into a multimodal macro model, and obtains a first judgment conclusion; after extracting text from all of the images, inputs the text and the first prompt word into a semantic macro model to obtain a second judgment conclusion; based on the first judgment conclusion and the second judgment conclusion, determines and outputs a final conclusion on the relevance of the review-related information and the professional title review materials. The present invention can effectively improve audit efficiency by using two models to judge the relevance of the review-related information and the professional title review materials, and determining the relevance of the review-related information and the professional title review materials based on the output results of the two models. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flow chart of a preferred embodiment of the method for reviewing professional title evaluation materials in the present invention.

[0045] Figure 2 It is a data processing flow chart of the present invention.

[0046] Figure 3 It is a structural diagram of a preferred embodiment of the professional title evaluation materials review device in the present invention.

[0047] Figure 4 It is a block diagram of the terminal principle of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the present invention more clear and distinct, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0049] In existing technology, applicants for professional title evaluations are required to select a series of supporting materials, such as papers, patents, projects, and awards, on the application form and upload the corresponding documents for review. However, there is currently no effective way to ensure that the information entered by applicants on the form is completely consistent with the uploaded attachments, and manual verification is still required. This manual verification method is inefficient.

[0050] In response to the above-mentioned defects of the prior art, the present invention provides a method, device, terminal and storage medium for reviewing professional title evaluation materials, which belongs to the field of review and certification technology. The method includes: receiving a number of review-related information and corresponding professional title evaluation materials; generating a first prompt word based on all the review-related information; pre-processing all the professional title evaluation materials to obtain a number of images, inputting all the images and the first prompt word into a multimodal large model to obtain a first judgment conclusion; after extracting text from all the images, inputting the text and the first prompt word into a semantic large model to obtain a second judgment conclusion; based on the first judgment conclusion and the second judgment conclusion, determining and outputting the final conclusion on the relevance of the review-related information and the professional title evaluation materials. The present invention can effectively improve the efficiency of review by using two models to judge the relevance of the review-related information and the professional title evaluation materials, and determining the relevance of the review-related information and the professional title evaluation materials based on the output results of the two models.

[0051] See Figure 1 The method for reviewing professional title evaluation materials according to an embodiment of the present invention includes the following steps:

[0052] Step S100: receiving several evaluation-related information input by the user in the professional title evaluation system interface, as well as corresponding professional title evaluation materials uploaded by the user for each evaluation-related information.

[0053] Specifically, when a user is conducting a professional title review, he or she needs to manually input different review-related information in the professional title review interface and upload the professional title review materials corresponding to the review-related information.

[0054] See Figure 1The method for reviewing professional title evaluation materials according to the embodiment of the present invention further includes the following steps:

[0055] Step S200: Generate a first prompt word based on all the review-related information.

[0056] Specifically, the first prompt word can inform the multimodal large model and the language large model of the review-related information that they need to pay attention to, and guide the model to specifically extract review-related content from the numerous information contained in the pictures or texts, and judge its relevance.

[0057] In one implementation, each review-related information includes the type of professional title review materials and the corresponding inspection items; and generating the first prompt word based on all the review-related information includes:

[0058] Pre-set the corresponding inspection items for each type of professional title review materials, and configure the corresponding prompt word template for each inspection item;

[0059] Determine all corresponding inspection items according to the type of the professional title evaluation materials input by the user;

[0060] Obtain the prompt word template corresponding to each inspection item, and fill the relevant inspection item content entered by the user into the corresponding prompt word template;

[0061] When all the inspection items are filled in the corresponding prompt word template, a first prompt word is generated based on the filled prompt word template.

[0062] Specifically, each review-related information includes the type of professional title review materials and the corresponding inspection items. By pre-establishing a correspondence between the types of professional title review materials and the corresponding inspection items, when the user selects the type of review material in the professional title review system interface, the corresponding inspection items will automatically appear. The user can enter the corresponding inspection items in the inspection item input box. For example, if the professional title review material type is a paper, the corresponding inspection items include the paper title, the name of the journal in which the paper is published, the journal level of the journal in which the paper is published, the year of publication, the research field of the paper, the number of citations, and the author information. The user can enter the information manually. In addition, a corresponding prompt word template is configured for each inspection item. For example, the prompt word template for the paper title is: "Please determine whether [ ] is the title of the paper in the image." The content between [ ] is the replaceable content of the module. When the user enters the paper title "Scenario Prediction Method Based on Prior Distribution" in the professional title review system interface, this is filled in the corresponding prompt word template, resulting in a prompt word template: "Please determine whether [Scenario Prediction Method Based on Prior Distribution] is the title of the paper in the image." Subsequently, when all the inspection items are filled in the corresponding prompt word template, a first prompt word is generated based on the filled prompt word template.

[0063] See Figure 1 The method for reviewing professional title evaluation materials according to the embodiment of the present invention further includes the following steps:

[0064] Step S300: pre-process all the professional title review materials to obtain a number of pictures, input all the pictures and the first prompt word into a multimodal large model, and after processing by the multimodal large model, output a first judgment conclusion on the relevance between the review-related information and the professional title review materials.

[0065] Specifically, the present invention uniformly processes different types of professional title review materials into image format, allowing the multimodal large model to process the materials in a relatively consistent manner. A multimodal large model refers to an artificial intelligence model capable of processing information in multiple modalities. Modality refers to the source or form of information, such as text, image, audio, video, etc. The present invention can invoke its image processing capabilities through the first prompt word.

[0066] In one implementation, the multimodal large model is any one of the GPT-4o and Tongyi Qianwen models.

[0067] In one implementation, all the professional title evaluation materials are pre-processed to obtain several images, including:

[0068] Determine the format of all the professional title evaluation materials and distinguish between image format files and non-image format files;

[0069] For professional title evaluation materials that are not in image format, convert them into images;

[0070] For professional title evaluation materials in image format, determine whether their resolution reaches the preset resolution;

[0071] If the preset resolution is not reached, a re-upload instruction including a description of the resolution requirement is generated, and a new image submitted by the user in response to the re-upload instruction is received, and the determination of whether the resolution of the new image reaches the preset resolution is continued until an image that reaches the preset resolution is obtained;

[0072] The pictures that have reached the preset resolution and the pictures that have undergone format conversion are combined to form the plurality of pictures.

[0073] Specifically, the professional title review materials uploaded by users include PDF and image formats. After receiving the professional title review materials uploaded by users, the format is first determined, and the professional title review materials in non-image format (i.e., professional title review materials in PDF format) are converted into images. For professional title review materials that are already in image format, it is determined whether the resolution reaches the preset resolution. Only when the preset resolution is reached can subsequent processing be continued. By performing format determination on professional title review materials in image format, the accuracy of information extraction in subsequent professional title reviews can be effectively ensured, thereby ensuring the accuracy of the final conclusion.

[0074] In one implementation, for professional title evaluation materials that are not in image format, converting them into images includes:

[0075] Determining whether the number of pages of each non-image format professional title evaluation material exceeds a preset number of pages;

[0076] If the number of pages of the professional title evaluation materials in non-picture format does not exceed the preset number of pages, the format of the materials is converted to obtain the corresponding pictures;

[0077] If the number of pages of the professional title evaluation materials in non-picture format exceeds the preset number of pages, they will be cut according to the preset number of pages, and the cut professional title evaluation materials in non-picture format will be format-converted to obtain corresponding pictures.

[0078] Specifically, to effectively control the amount of data input into a large multimodal model, a preset number of pages is set. Materials exceeding the preset number of pages are truncated. This approach ensures faster model processing.

[0079] In one implementation, the preset number of pages ranges from 5 to 30 pages.

[0080] See Figure 1 The method for reviewing professional title evaluation materials according to the embodiment of the present invention further includes the following steps:

[0081] Step S400: After extracting text from all the images, input the text and the first prompt word into a semantic big model, and the semantic big model processes and outputs a second judgment conclusion on the relevance of the review-related information and the professional title review materials.

[0082] Specifically, in addition to utilizing a multimodal large language model to determine the relevance of review-related information to professional title review materials, the present invention also utilizes a large language model for judgment. After extracting text from all of the images using optical character recognition (OCR) technology, the text and the first prompt word are input into a large semantic model for judgment. The large semantic model is an artificial intelligence model that primarily focuses on understanding and generating semantic information in natural language. By studying large amounts of text and other language data, it can uncover semantic relationships between words, sentences, and paragraphs in a language, thereby enabling numerous natural language processing tasks such as text classification, sentiment analysis, question-answering systems, and machine translation.

[0083] In one implementation, the semantic big model is any one of a BERT model (Bidirectional Encoder Representations from Transformers), an ERNIE (Enhanced Representation through kNowledge IntEgration) model, and an XLNet model.

[0084] See Figure 1 The method for reviewing professional title evaluation materials according to the embodiment of the present invention further includes the following steps:

[0085] Step S500: Based on the first judgment conclusion and the second judgment conclusion, determine and output the final conclusion on the relevance of the review-related information and the professional title review materials.

[0086] Specifically, optical character recognition (OCR) technology can extract text from images. However, for professional title review materials, using only OCR to extract text from images fails to effectively process the original format and order of the materials, resulting in inaccurate information extraction. Using only a single multimodal model or semantic large model for processing can also lead to biased judgments due to their differing focus. The present invention combines the judgment conclusions output by two models to determine and output a final conclusion regarding the relevance of review-related information to the professional title review materials. This approach effectively leverages the strengths of both models, resulting in a more accurate final conclusion. Specifically, a determination is made as to whether the first and second judgment conclusions are consistent; if they are consistent, either the first or second judgment conclusion is output as the final conclusion. Because the two models focus on different aspects of processing, the multimodal large model can comprehensively consider multiple aspects of image information, such as text, graphics, and format, leveraging its ability to perceive the overall image to determine relevance. The semantic large model extracts text based on the picture and combines it with the first prompt word to draw a second judgment conclusion, focusing on in-depth understanding of the text semantics. When the two judgment conclusions are consistent, the present invention will use the first judgment conclusion or the second judgment conclusion as the final conclusion and output it, which can effectively avoid the problem of inaccurate judgment. If the two are inconsistent, a second prompt word is constructed, and the picture and the second prompt word are input into the multimodal large model, and the final conclusion is output after processing. This move can further mine the picture information, make the final conclusion more in line with the actual situation, and enhance the adaptability to complex and diverse professional title evaluation materials.

[0087] In one implementation, the multimodal big model outputs a first reasoning process while outputting a first judgment conclusion. The first reasoning process is the process of the multimodal big model reasoning the relevance between the review-related information and the professional title review materials. The semantic big model outputs a second reasoning process while outputting a second judgment conclusion. The second reasoning process is the process of the semantic big model reasoning the relevance between the review-related information and the professional title review materials.

[0088] Specifically, the multimodal large model can simultaneously output the first judgment conclusion and the first reasoning process, and the semantic large model can simultaneously output the second judgment conclusion and the second reasoning process.

[0089] In one implementation, the step of constructing the second prompt word includes:

[0090] Obtaining a preset second prompt word template, where the second prompt word template is used for the multimodal large model to perform a two-choice judgment;

[0091] The first prompt word, the first judgment conclusion, the first reasoning process, the second judgment conclusion and the second reasoning process are filled into the second prompt word template to obtain a second prompt word.

[0092] Specifically, the first prompt word, first judgment conclusion, first reasoning process, second judgment conclusion, and second reasoning process are entered into the second prompt word template. This allows the multimodal large model to fully consider all previous relevant information when re-judging, and then make a binary judgment, selecting either the first judgment conclusion or the second judgment conclusion as the final conclusion and outputting it. This helps improve the quality of the final conclusion, making relevance judgments more accurate and reasonable, and better able to cope with information conflicts and uncertainties when dealing with complex review materials.

[0093] The data processing flow chart of the present invention is as follows: Figure 2 As shown. The user first enters the type of materials and the content of the inspection items in the form on the professional title evaluation system interface, and uploads the corresponding professional title evaluation materials. A first prompt word can be generated based on the input content. The professional title evaluation materials are preprocessed into several pictures. The pictures and the first prompt word are input into the multimodal large model to obtain the first judgment result and the first reasoning process. After optical character recognition (OCR) of the picture, text is obtained. The text and the first prompt word are input into the semantic large model to obtain the second judgment conclusion and the second reasoning process. Determine whether the two judgment results are consistent. If they are consistent, either judgment result is used as the final result. If they are inconsistent, a second prompt word is constructed, and the picture and the second prompt word are input into the multimodal large model for a two-choice judgment to output the final conclusion.

[0094] In summary, the present invention receives a number of review-related information and corresponding professional title review materials; generates a first prompt word based on all the review-related information; pre-processes all the professional title review materials to obtain a number of images, inputs all the images and the first prompt word into a multimodal large model, and obtains a first judgment conclusion; after extracting text from all the images, inputs the text and the first prompt word into a semantic large model to obtain a second judgment conclusion; based on the first judgment conclusion and the second judgment conclusion, determines and outputs the final conclusion on the relevance of the review-related information and the professional title review materials. The present invention can effectively improve the audit efficiency by using two models to judge the relevance of the review-related information and the professional title review materials, and determines the relevance of the review-related information and the professional title review materials based on the output results of the two models.

[0095] In one embodiment, if Figure 3 As shown, based on the above-mentioned method for reviewing professional title evaluation materials, the present invention also provides a professional title evaluation materials review device, comprising:

[0096] The data receiving module 100 is used to receive a number of review-related information input by the user in the professional title review system interface, as well as the corresponding professional title review materials uploaded by the user for each review-related information;

[0097] A prompt word generating module 200 is used to generate a first prompt word based on all the review-related information;

[0098] A first judgment module 300 is configured to pre-process all of the professional title evaluation materials to obtain a plurality of images, input all of the images and the first prompt word into a multimodal large model, and output a first judgment conclusion regarding the relevance of the evaluation-related information and the professional title evaluation materials after processing by the multimodal large model;

[0099] The second judgment module 400 is configured to extract text from all the images, input the text and the first prompt word into a semantic macro model, process the semantic macro model, and output a second judgment conclusion on the relevance of the review-related information to the professional title review materials;

[0100] The final conclusion generating module 500 is used to determine and output the final conclusion on the relevance between the review-related information and the professional title review materials based on the first judgment conclusion and the second judgment conclusion.

[0101] In one embodiment, each review-related information includes the type of professional title review materials and the corresponding inspection items; the prompt word generation module includes:

[0102] The preset unit is used to pre-set the inspection items corresponding to each type of professional title evaluation materials and configure the corresponding prompt word template for each inspection item;

[0103] An inspection item determination unit, configured to determine all corresponding inspection items according to the type of the professional title evaluation materials input by the user;

[0104] An information filling unit is used to obtain a prompt word template corresponding to each inspection item and fill the relevant inspection item content input by the user into the corresponding prompt word template;

[0105] The information integration unit is configured to generate a first prompt word based on the filled prompt word template when all the inspection item contents are filled in the corresponding prompt word template.

[0106] In one embodiment, the apparatus further comprises:

[0107] A format judgment unit, used to judge the format of all the professional title evaluation materials and distinguish between image format files and non-image format files;

[0108] Image conversion unit, used for converting the professional title evaluation materials that are not in image format into images;

[0109] A resolution judgment unit is used to judge whether the resolution of the professional title evaluation materials in the image format reaches the preset resolution;

[0110] a re-upload unit, configured to generate a re-upload instruction including a resolution requirement description if the preset resolution is not reached, receive a new image submitted by the user in response to the re-upload instruction, and continue to determine whether the resolution of the new image reaches the preset resolution until an image that reaches the preset resolution is obtained;

[0111] The plurality of picture generating units are used to combine pictures having a preset resolution and pictures having undergone format conversion to form the plurality of pictures.

[0112] In one embodiment, the apparatus further comprises:

[0113] The first judging unit is configured to judge whether the number of pages of each non-picture format professional title evaluation material exceeds a preset number of pages;

[0114] A first format conversion unit is configured to convert the professional title evaluation materials in non-picture format into corresponding pictures if the number of pages does not exceed a preset number of pages;

[0115] The second format conversion unit is used to cut the professional title evaluation materials in non-picture format according to the preset number of pages if the number of pages exceeds the preset number of pages, and convert the cut professional title evaluation materials in non-picture format to obtain corresponding pictures.

[0116] In one embodiment, the final conclusion generation module includes:

[0117] A second judgment unit is used to judge whether the first judgment conclusion and the second judgment conclusion are consistent;

[0118] A first conclusion output unit, configured to output the first judgment conclusion or the second judgment conclusion as a final conclusion if the two are consistent;

[0119] The second conclusion output unit is used to construct a second prompt word if the two are inconsistent, input the picture and the second prompt word into the multimodal large model, and output the final conclusion after processing.

[0120] In one embodiment, the apparatus further comprises:

[0121] A template acquisition unit, configured to acquire a preset second prompt word template, wherein the second prompt word template is used for performing a one-or-nothing judgment in the multimodal large model;

[0122] The second prompt word generating unit is configured to fill the first prompt word, the first judgment conclusion, the first reasoning process, the second judgment conclusion and the second reasoning process into the second prompt word template to obtain a second prompt word.

[0123] Based on the above embodiment, the present invention further provides a terminal, whose principle block diagram can be shown as follows: Figure 4 As shown. The terminal includes a processor, a memory, a network interface and a display screen connected via a device bus. The processor of the terminal is used to provide computing and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating device and a professional title evaluation materials review program. The internal memory provides an environment for the operation of the operating device and the professional title evaluation materials review program in the non-volatile storage medium. The network interface of the terminal is used to communicate with an external terminal via a network connection. When the professional title evaluation materials review program is executed by the processor, the steps of any one of the above-mentioned professional title evaluation materials review methods are implemented. The display screen of the terminal can be a liquid crystal display or an electronic ink display.

[0124] Those skilled in the art will understand that Figure 4 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0125] In one embodiment, a terminal is provided, which includes a memory, a processor, and a professional title evaluation materials review program stored in the memory and runnable on the processor. When the professional title evaluation materials review program is executed by the processor, the steps of any one of the professional title evaluation materials review methods provided in the embodiments of the present invention are implemented.

[0126] An embodiment of the present invention also provides a computer-readable storage medium, on which a professional title evaluation materials review program is stored. When the professional title evaluation materials review program is executed by a processor, the steps of any one of the professional title evaluation materials review methods provided in an embodiment of the present invention are implemented.

[0127] It should be understood that the sequence numbers of the steps in the above embodiments do not imply a specific order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0128] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the above-mentioned device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned device can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0129] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0130] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0131] In the embodiments provided by the present invention, it should be understood that the disclosed apparatus / terminal device and method can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For example, the division of the modules or units described above is merely a logical functional division. In actual implementation, other division methods may be used. For example, multiple units or components may be combined or integrated into another device, or some features may be omitted or not implemented.

[0132] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for reviewing professional title evaluation materials, characterized in that: The method comprises: Receive several review-related information entered by the user in the professional title review system interface, as well as the corresponding professional title review materials uploaded by the user for each review-related information. Each review-related information includes the type of professional title review materials and the corresponding inspection items. Generate a first prompt word based on all the review-related information, where the first prompt word is used to guide the model to specifically extract review-related content from a large amount of information contained in the image or text; Preprocessing all of the professional title review materials to obtain a plurality of images, inputting all of the images and the first prompt word into a multimodal large model, and processing the multimodal large model to output a first judgment conclusion and a first reasoning process regarding the relevance between the review-related information and the professional title review materials, wherein the first reasoning process is a process of inferring the relevance by the multimodal large model; After extracting text from all the images, the text and the first prompt word are input into a semantic macro model. After processing by the semantic macro model, a second judgment conclusion and a second reasoning process regarding the relevance between the review-related information and the professional title review materials are output. The second reasoning process is a process in which the semantic macro model infers the relevance. Determining whether the first judgment conclusion and the second judgment conclusion are consistent; If the two are consistent, the first judgment conclusion or the second judgment conclusion is taken as the final conclusion and output; If the two are inconsistent, obtain a preset second prompt word template. The second prompt word template is used for the multimodal large model to make a two-choice judgment. The first prompt word, the first judgment conclusion, the first reasoning process, the second judgment conclusion and the second reasoning process are filled in the second prompt word template to obtain the second prompt word. The picture and the second prompt word are input into the multimodal large model, and the final conclusion is output after processing.

2. The method for reviewing professional title evaluation materials according to claim 1, characterized in that: Based on all the review-related information, generating the first prompt word includes: Pre-set the corresponding inspection items for each type of professional title review materials, and configure the corresponding prompt word template for each inspection item; Determine all corresponding inspection items according to the type of the professional title evaluation materials input by the user; Obtain the prompt word template corresponding to each inspection item, and fill the relevant inspection item content entered by the user into the corresponding prompt word template; When all the inspection items are filled in the corresponding prompt word template, a first prompt word is generated based on the filled prompt word template.

3. The method for reviewing professional title evaluation materials according to claim 1, characterized in that: All the professional title evaluation materials are pre-processed to obtain several pictures, including: Determine the format of all the professional title evaluation materials and distinguish between image format files and non-image format files; For professional title evaluation materials that are not in image format, convert them into images; For professional title evaluation materials in image format, determine whether their resolution reaches the preset resolution; If the preset resolution is not reached, a re-upload instruction including a description of the resolution requirement is generated, and a new image submitted by the user in response to the re-upload instruction is received, and the determination of whether the resolution of the new image reaches the preset resolution is continued until an image that reaches the preset resolution is obtained; The pictures that have reached the preset resolution and the pictures that have undergone format conversion are combined to form the plurality of pictures.

4. The method for reviewing professional title evaluation materials according to claim 3, characterized in that: For professional title evaluation materials that are not in image format, convert them into images, including: Determining whether the number of pages of each non-image format professional title evaluation material exceeds a preset number of pages; If the number of pages of the professional title evaluation materials in non-picture format does not exceed the preset number of pages, the format of the materials is converted to obtain the corresponding pictures; If the number of pages of the professional title evaluation materials in non-picture format exceeds the preset number of pages, they will be cut according to the preset number of pages, and the cut professional title evaluation materials in non-picture format will be format-converted to obtain corresponding pictures.

5. A device for reviewing professional title evaluation materials, characterized in that: include: The data receiving module is used to receive several review-related information input by the user in the professional title review system interface, as well as the corresponding professional title review materials uploaded by the user for each review-related information. Each review-related information includes the type of professional title review materials and the corresponding inspection items. A prompt word generation module is used to generate a first prompt word based on all the review-related information, wherein the first prompt word is used to guide the model to specifically extract content related to the review from a large amount of information contained in the image or text; A first judgment module is configured to pre-process all of the professional title review materials to obtain a plurality of images, input all of the images and the first prompt word into a multimodal large model, and output a first judgment conclusion and a first reasoning process regarding the relevance between the review-related information and the professional title review materials after processing by the multimodal large model, wherein the first reasoning process is a process of inferring the relevance by the multimodal large model; A second judgment module is configured to extract text from all the images, input the text and the first prompt word into a semantic macro model, process the semantic macro model, and output a second judgment conclusion and a second reasoning process regarding the relevance between the review-related information and the professional title review materials, wherein the second reasoning process is a process of reasoning the relevance using the semantic macro model; A final conclusion generating module, configured to determine whether the first judgment conclusion and the second judgment conclusion are consistent; If the two are consistent, the first judgment conclusion or the second judgment conclusion is taken as the final conclusion and output; If the two are inconsistent, obtain a preset second prompt word template. The second prompt word template is used for the multimodal large model to make a two-choice judgment. The first prompt word, the first judgment conclusion, the first reasoning process, the second judgment conclusion and the second reasoning process are filled in the second prompt word template to obtain the second prompt word. The picture and the second prompt word are input into the multimodal large model, and the final conclusion is output after processing.

6. A terminal, characterized in that: The terminal includes a memory, a processor, and a professional title evaluation material review program stored in the memory and runnable on the processor. When the professional title evaluation material review program is executed by the processor, the steps of the professional title evaluation material review method as described in any one of claims 1-4 are implemented.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a professional title evaluation materials review program. When the professional title evaluation materials review program is executed by the processor, the steps of the professional title evaluation materials review method according to any one of claims 1 to 4 are implemented.

Images